paradiseo/doc/edo/index.h
Adèle Harrissart 490e837f7a * New tree configuration of the project:
.../
   ...           + -- EO
   |             |
   |             |
   +-- src ----- + -- EDO
   |             |
   |             |
   +-- test      + -- MO
   |             |
   |             |
   +-- tutorial  + -- MOEO
   |             |
   |             |
   +-- doc       + -- SMP
   |             |
   |             |
   ...           + -- EOMPI
                 |
                 |
                 + -- EOSERIAL

Question for current maintainers: ./README: new release?

Also:

* Moving out eompi & eoserial modules (issue #2).

* Correction of the errors when executing "make doc" command.

* Adding a solution for the conflicting headers problem (see the two CMake Cache
 Values: PROJECT_TAG & PROJECT_HRS_INSTALL_SUBPATH) (issue #1)

* Header inclusions:
        ** src: changing absolute paths into relative paths ('#include <...>' -> '#include "..."')
        ** test, tutorial: changing relative paths into absolute paths ('#include "..."' -> '#include <...>')

* Moving out some scripts from EDO -> to the root

* Add a new script for compilation and installation (see build_gcc_linux_install)

* Compilation with uBLAS library or EDO module: now ok

* Minor modifications on README & INSTALL files

* Comment eompi failed tests with no end

*** TODO: CPack (debian (DEB) & RedHat (RPM) packages) (issues #6 & #7) ***
2014-09-06 13:04:35 +02:00

39 lines
1.5 KiB
C
Executable file

/** @mainpage Welcome to Evolving Distribution Objects
@section shortcuts In one word
%EDO is an extension of %EO oriented toward Estimation-of-Distribution-like
Algorithms.
You can search for advanced features by browsing the <a
href="modules.html">modules</a> page.
@section intro Introduction
%EDO is an extension of %EO, that facilitate the design and implementation of
stochastic search metaheuristics. It is based on the assumption that those
algorithms are updating a probability distribution, that is used to generate
a sample (a population, in %EO) of solutions (individuals, in %EO).
Basically, EDO decompose the <em>variation</em> operators of %EO in a set of
sub-operators that are binded by a <em>distribution</em>. Thus, most of the
representation-independent operators of %EO can be used in %EDO algorithms.
Apart from choosing which distribution he want to use as a model, the user is
not supposed to directly manipulate it. Using the same approach than within %EO,
the user has just to indicate what he want to use, without having to bother how
he want to use it.
On the designer side, it is still possible to implement specific operators
without having to change other ones.
<img src="edo_design.png" />
The two main operators are the <em>Estimators</em>, that builds a given
distribution according to a population and the <em>Samplers</em> that builds a
population according to a distribution. There is also <em>Modifiers</em> that
are here to change arbitrarily the parameters of a distribution, if necessary.
<img src="edo_distrib.png" />
*/